Testing smoothing and sinuosity

Loading packages

list_of_packages <- c("tidyverse", "plotly", "here", "signal", "DT")
new_packages <- list_of_packages[!(list_of_packages %in% installed.packages()[,"Package"])]
if(length(new_packages)) install.packages(new_packages)

library(tidyverse)

## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.0 ──

## ✓ ggplot2 3.3.3     ✓ purrr   0.3.4
## ✓ tibble  3.1.0     ✓ dplyr   1.0.4
## ✓ tidyr   1.1.2     ✓ stringr 1.4.0
## ✓ readr   1.4.0     ✓ forcats 0.5.1

## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

library(plotly)

## 
## Attaching package: 'plotly'

## The following object is masked from 'package:ggplot2':
## 
##     last_plot

## The following object is masked from 'package:stats':
## 
##     filter

## The following object is masked from 'package:graphics':
## 
##     layout

library(here)

## here() starts at /home/francisko/coding/r/flight-path-experiment

library(signal)

## 
## Attaching package: 'signal'

## The following object is masked from 'package:plotly':
## 
##     filter

## The following object is masked from 'package:dplyr':
## 
##     filter

## The following objects are masked from 'package:stats':
## 
##     filter, poly

library(DT)

Loading data

b195_t1 <- readr::read_csv(here::here("unpaired-points", "b195-t1-unpaired-points-xyz.csv")) 

## 
## ── Column specification ────────────────────────────────────────────────────────
## cols(
##   x_1 = col_double(),
##   y_1 = col_double(),
##   z_1 = col_double()
## )

Testing smoothing

Creating filter

I created the filter using the same parameter values Camille describes in his paper.

butterworth_filter <-
  butter(n = 4,  # order
         W = 0.5,  # critical frequency
         type = "low")

Applying filter

b195_t1 %>% 
  tidyr::drop_na() %>% 
  dplyr::mutate(x_butterworth = signal::filter(butterworth_filter, x_1),
                y_butterworth = signal:: filter(butterworth_filter, y_1),
                z_butterworth = signal::filter(butterworth_filter, z_1)) -> b195_t1_filtered

b195_t1_filtered %>% 
  DT::datatable(extensions = 'Buttons',
                options = list(dom = 'Blfrtip',
                               buttons = c('copy', 'csv', 'excel', 'pdf', 'print'),
                               lengthMenu = list(c(10,25,50,-1),
                                                 c(10,25,50,"All"))))

Here is a plot with both the raw and the “smoothed” paths so you can see difference.

plot_ly(data = b195_t1_filtered,
        type = "scatter3d",
        name = "Raw tracking",
        x = b195_t1_filtered$x_1,
        y = b195_t1_filtered$y_1,
        z = b195_t1_filtered$z_1,
        mode = "lines",
        line = list(color = "steelblue")) %>% 
  add_trace(
    type = "scatter3d",
    name = "Tracking smoothed usind butterworth filter",
    x = b195_t1_filtered$x_butterworth,
    y = b195_t1_filtered$y_butterworth,
    z = b195_t1_filtered$z_butterworth,
    mode = "lines",
    line = list(color = "firebrick")
  )

Testing sinuosity

A quick intro to how I calculated the distance the butterflies flew (I dare you to say it three times fast!)

I used dist(), a function from base R. It uses Euclidean distances by default and, since I did not change it, that’s the distances I sent you in that other Rmd file.

I’ll give a brief tour of how I calculated the distances to clarify things.

I defined the following function to calculate distance, it:

• removes NAs, which is required for the next steps;
• calls dist to calculate the distances;
• converts it into a matrix;
• adds the distances between the points (distance between points 1 and 2, points 2 and 3, and so on);

calc_flight_dist <- name <- function(dat) {
  dat_dist <-  {{dat}} %>% 
  dplyr::filter(!is.na(x_1)) %>%  # removes NA
  dist() %>%  # calculates distance
  as.matrix()  # converts it to a matrix

return(sum((dat_dist[row(dat_dist) == col(dat_dist) + 1])))  # sums the distance between points
  
}

Here is it working in practice. First, I define a toy set of points. In this example there’s only movement along one axis to make it easier to follow

tibble(
  x_1 = c(0, 3, 5),  # there' only movement along this axis
  y = c(0, 0, 0),
  z = c(0, 0, 0)
) %>% 
  calc_flight_dist()

## [1] 5

dist_dat <- 
  b195_t1 %>% 
  dplyr::filter(!is.na(x_1)) %>%
  dist() %>% 
  as.matrix()

dist_dat[1, nrow(dist_dat)]

## [1] 2.124259